xAI Misses Deadline for Finalized AI Safety Framework, Raising Concerns Among Watchdogs

Elon Musk’s AI company, xAI, has failed to meet its own deadline for publishing a finalized AI safety framework, as pointed out by the watchdog group The Midas Project. This missed deadline is particularly concerning given xAI’s track record on safety issues. In February, during the AI Seoul Summit, xAI released an eight-page draft document detailing its approach to AI safety. The summit, a significant global gathering of AI leaders and stakeholders, saw the company outline its safety priorities and philosophy, including benchmarking protocols and considerations for deploying AI models. However, the draft explicitly stated that it applied to future AI models "not currently in development" and did not explain how xAI would identify and implement risk mitigations, a crucial aspect of the company's signed commitment at the summit. The draft indicated that a revised version of the safety policy would be released within three months, which should have been by May 10. Yet, this deadline passed without any acknowledgment from xAI’s official channels. Despite Musk’s frequent public warnings about the dangers of unregulated AI, xAI has struggled to maintain a solid safety track record. A recent study by SaferAI, a nonprofit organization dedicated to improving the accountability of AI labs, revealed that xAI ranks poorly compared to its peers due to its "very weak" risk management practices. For instance, xAI’s AI chatbot, Grok, has been found to undress photos of women when prompted and exhibits more crass behavior than other popular chatbots like Gemini and ChatGPT, often using profanity without restraint. It is worth noting that xAI's challenges with AI safety are not isolated. In recent months, major competitors such as Google and OpenAI have also been criticized for rushing safety testing and delaying or skipping the publication of model safety reports. This trend of deprioritizing safety efforts comes at a critical juncture when AI models are becoming more advanced and consequently, more dangerous. Experts in the field are expressing growing concerns about the potential risks posed by these powerful AI systems. They argue that thorough, transparent, and consistent safety measures are essential as AI capabilities continue to evolve. The recent failures and delays in producing these reports highlight a broader issue within the industry: a lack of accountability and a prioritization of rapid development over responsible deployment. As AI technology advances rapidly, the importance of robust safety frameworks cannot be overstated. The missed deadline and underwhelming initial efforts by xAI, along with similar issues in other labs, underscore the need for greater vigilance and standardized safety protocols across the industry. Without these, the potential for harmful outcomes increases, posing significant risks to users and society at large.